Research Design

This study utilized a cross-sectional design for the collection of data at one specific time. Cross-sectional studies involve collecting data one time and analyzing that information to examine how variables correlate with each other. The objective of this particular study was to assess the relationship between workplace culture, burnout experienced by nurses, and nurse’s intent to continue working.

Although cross-sectional designs do limit researchers from determining cause-and-effect relationships, the use of cross-sectional designs in organizational health research is common for testing theoretical indirect pathways initially (such as those indicated by the JD-R model), and in this case, testing if workplace culture functions as a job resource to reduce job demand on burnout which would then influence withdrawal cognition (the nurse’s decision to leave their employment). Since this study has a single time-point design, it can be used to analyze the pattern of covariance among variables and establish if the covariance is consistent with the proposed mediation process. However, due to its association nature instead of being causal in nature, the directionality of the paths must be interpreted appropriately. Temporal precedence and causation require further longitudinal or experimental designs.

Setting and Participants

The research was conducted in the Hail and Qassim regions of Saudi Arabia. The study locations specifically included King Khalid Hospital, King Salman Specialist Hospital, and Hail General Hospital in the Hail region, alongside Qassim Medical Centre and King Saud Hospital in the Qassim region. Participants were full-time registered nurses (RNs) working directly with patients who had completed their probationary period. The sample included 41.1% male nurses, which is higher than typical Saudi nursing workforce distributions. This reflects the specific hospital settings sampled and should be considered when generalizing findings to the broader Saudi nursing population, which remains predominantly female. Nurses on extended maternity or study leave, or those in purely administrative roles, were excluded.

These hospitals were purposively selected to represent the diversity of public healthcare delivery in the region. The sample includes general tertiary hospitals (King Khalid Hospital, Hail General Hospital, King Saud Hospital), a specialist tertiary center (King Salman Specialist Hospital), and a large integrated medical city (Qassim Medical Centre). Bed capacities range from approximately 300–800 beds, and the facilities collectively cover the full spectrum of inpatient and outpatient services, from primary to quaternary care. This variation in institutional size, scope, and patient acuity was intended to enhance external validity by ensuring that findings are not confined to a single type of healthcare setting. All selected hospitals operate under the Ministry of Health public health cluster system, which is the dominant employer of nurses in these regions.

Sampling Methodology and Sample Size Calculations

A methodologically rigorous approach was used to select an appropriately representative sample of nurses from each of the Primary Health Care Centers listed above. A proportionate stratified random sampling technique was employed. The sample was divided into two distinct geographic strata; namely the Hail Region (three hospitals) and the Qassim Region (two hospitals). Overall, the population comprised 2500 nurses. Using a computer generated random number, 400 nurses were randomly selected proportional to their representation; 240 from the Hail Region and 160 from the Qassim Region.

Prior to conducting the study, a priori power calculations were conducted to determine the required sample size for the planned mediation SEM. Using Monte Carlo power analysis for indirect effects [23], with anticipated path coefficients of a = −0.30 (culture → burnout), b = −0.30 (burnout → intent to stay), c’ = 0.45 (direct effect), and based on 5,000 simulations, a sample size of 300 was estimated to achieve 80% power to detect the indirect effect at α = 0.05. As a conservative check, power for Pearson correlation (r = 0.16) was also calculated, yielding a minimum n = 305. Accounting for an estimated 20% non-response rate, the recruitment target was set at 400 nurses. The final sample of 355 exceeded both power thresholds.

The final sample size of 355 is also adequate for the planned structural equation modeling (SEM) analysis. With 57 observed items and 21 estimated parameters in the measurement model, plus 6 structural paths in the mediation model, the ratio of sample size to estimated parameters exceeds 5:1, satisfying minimum recommendations for SEM. Proportionate allocation across the two regional strata (Hail: 67.6% of sample; Qassim: 32.4% of sample) mirrors the relative distribution of nurses across the two health clusters, supporting the generalizability of structural parameter estimates to the broader nursing population in these regions.

Instrumentation

Three validated psychometric instruments were used in this research. First, the Organizational Culture Survey [24] utilized a 31-item scale with a 5-point Likert-type response scale (1 = “to a very little extent” to 5 = “to a very large extent”) to measure the perception of nurses regarding workplace culture.

Second, the Copenhagen Burnout Inventory (CBI) [9] consisted of a 19-item scale assessing personal, work-related, and client-related burnout. The CBI utilized a 5-point frequency scale (1 = “never” to 5 = “always”) where an increased mean score represents increased perceived burnout.

Third, the Intent to Stay Scale [25] featured a 7-item scale utilizing a 5-point agreement scale (1 = “strongly disagree” to 5 = “strongly agree”), with items 2, 3, 5, and 7 being reverse-coded. Expert review by three specialists was utilized to confirm the content validity of these measures. A pilot study was conducted with 15 nurses who were not included in the main sample to assess the clarity of items, response time, and preliminary internal consistency. Based on pilot feedback, minor wording adjustments were made to three items for cultural appropriateness. Cronbach’s alpha coefficient values from the pilot were.93 for the Organizational Culture Survey,.89 for the Copenhagen Burnout Inventory, and.79 for the Intent to Stay Scale.

Data Collection Process

Formal approvals were obtained from both the Health Clusters from the Hail region prior to conducting the research. Participants were identified via nursing administration and HR databases. Facilitators from each department were then contacted to forward a Google Forms link to the selected nurses via official email and professional WhatsApp groups. Data collection took place over a period of approximately two months, starting on November 15, 2025, and concluding on January 15, 2026.

Ethical Considerations

All aspects of the research were conducted in accordance with the ethical standards established by the Health Cluster Institutional Review Board, Hail Region, Ministry of Health, Kingdom of Saudi Arabia (Approval No. #2024−093 dated October 23, 2024). Confidentiality and anonymity protections were strictly implemented, and no identifiers of personal information were included in the collected data. All participants received a written informed consent statement detailing the objective of the study, the protections to maintain participant confidentiality, and the voluntary nature of participation. All participants were informed that they could withdraw from the study at any time without penalty.

Statistical Analysis

In addition to descriptive statistics (means and SD), preliminary statistical analysis of the data consisted of reliability testing (Cronbach’s alpha). Bivariate correlation (Pearson’s r) analysis was also used to determine the initial linear relationship between the variables of interest. The primary analysis consisted of a mediation analysis model using structural equation modeling (SEM) to analyze the structural relationship from workplace culture to intent to stay through nurse burnout. The bootstrapping method with a 95 percent confidence interval was also utilized to validate the mediation effect through the examination of the indirect effect.

All analyses were conducted using Mplus version 8.8 [26]. Given the Likert-type nature of the items and likely deviation from multivariate normality, the robust maximum likelihood estimator (MLR) was employed, providing standard errors and a scaled chi-square statistic robust to non-normality. Missing data were minimal (< 2% across all items) and were handled using full information maximum likelihood (FIML), which uses all available data to estimate parameters without listwise deletion or imputation.

The organizational culture survey [24] is a six-factor instrument (Teamwork/Conflict, Climate/Morale, Information Flow, Involvement, Supervision, Meetings). For the present study, all 31 items were specified as indicators of a single second-order workplace culture latent variable, with the six subscales loading onto the higher-order factor. Similarly, the CBI comprises three established subscales (personal, work-related, client-related), modeled as first-order factors loading onto a single second-order Burnout latent variable. A unidimensional alternative was tested and produced inferior fit; the second-order specification was retained. No item parcels were created as all 57 items served as observed indicators at the first-order level.

Prior to testing the structural model, a confirmatory factor analysis (CFA) was conducted to evaluate the measurement model. Model fit was assessed using χ², CFI, TLI, RMSEA, and SRMR. Acceptable fit was defined by the following criteria: Comparative Fit Index (CFI) ≥ 0.90, Tucker-Lewis Index (TLI) ≥ 0.90, Root Mean Square Error of Approximation (RMSEA) ≤ 0.08 with 90% confidence interval upper bound < 0.08, and Standardized Root Mean Square Residual (SRMR) ≤ 0.08. Convergent validity was examined via standardized factor loadings (λ ≥ .50), composite reliability (ρc ≥ .70), and average variance extracted (AVE ≥ .50). Discriminant validity was assessed using the Fornell–Larcker criterion (AVE > squared correlation between constructs) and the heterotrait-monotrait (HTMT) ratio of correlations (HTMT < .85).

Because all variables were measured using self-report at a single time point, common method variance (CMV) was assessed using two approaches. First, Harman’s single-factor test was conducted by entering all items into an exploratory factor analysis (EFA) with one fixed factor. A single factor accounting for less than 50% of the total variance suggests that CMV is not a serious concern [27]. Second, an unmeasured latent method factor (ULMF) analysis was conducted by adding a latent method factor to the CFA model, with all items loading on this factor [27]. A substantial improvement in model fit (ΔCFI > .01) after adding the method factor would indicate significant CMV.

Relationship of workplace culture, burnout and intent to stay

Workplace culture, employee burnout, and employee retention are all among the most significant concerns for both organizational psychology and human resource management. The relationships identified were consistent with the JD-R model claims that job resources will mitigate strain and subsequently decrease withdrawal cognitions [19]. However, how workplace culture affects the organization may be more complex than simply a resource demand equilibrium. In Saudi tertiary hospitals, workplace culture may act not only as a protective mechanism against work demands, but also as a signal of the organization’s value placed upon their employees, thus strengthening normative commitment independent of reducing burnout.

A negative, moderate correlation was identified between burnout and the desire to remain in the workplace. Overall, the data collected indicate that a positive workplace culture contributes to employee retention; however, burnout may significantly detract from employee’s commitment to staying at the workplace.

There is considerable literature showing that organizations can proactively reduce employee burnout by developing a positive workplace environment. Literature shows that there exists an inverse relationship between a positive workplace culture (i.e., relational, innovative) and employee burnout. Conversely, there exists a positive relationship between a negative workplace culture (i.e., hierarchical), and employee burnout. Ultimately, this relationship impacts whether or not an employee decides to terminate employment with the organization [10]. The Job Demands-Resources (JD-R) Model indicates that job resources (i.e., positive workplace culture) can serve as a protective factor for employees against stressors. Moreover, workplace incivility has been strongly linked to both increased burnout and increased turnover intentions [19,37].

An employee’s intent to remain at an organization is influenced by many factors, including the level of organizational support experienced by the employee, the employee’s perceived job security, and the employee’s overall job satisfaction. As indicated in the current study, the positive correlation between job satisfaction and the intent to remain at an organization suggests that when employees are happy with their workplace, they are more likely to continue working at the organization. Previous studies have also reported that a high level of burnout increases an employee’s desire to leave [38,39]. In addition, the results suggest that the relationship between burnout and turnover intentions demonstrates that how employees are treated will greatly affect their engagement, and subsequent decisions about their employment. These results support the idea that burnout is not solely caused by a high level of job demands. Rather, they provide evidence that organizational culture functions as a protective mechanism against burnout and as a determinant of an employee’s engagement [40]. Although evidence has provided mixed results depending upon the cultural context or hierarchical structure [10] the central theme of all of the results is clear that positive workplace cultures promote retention. Therefore, it is crucial that organizational managers evaluate workplace culture, and develop and implement interventions aimed at reducing burnout and promoting nurse retention.

The present findings appear consistent with studies linking supportive cultures to lower burnout [1,11] yet this consistency should be interpreted cautiously. Several studies report null or positive associations between hierarchical culture and retention [10], suggesting that cultural effects may depend on how culture is operationalized. For example, cultures emphasizing stability and predictability—often coded as ‘hierarchical’—may reduce uncertainty and thereby lower burnout in high-stakes clinical environments, even though such cultures are typically viewed as restrictive [10]. Methodologically, these discrepancies may arise from differences in culture measurement: the organizational culture survey used here assesses perceived support and communication, whereas instruments classifying cultures as clan, adhocracy, market, or hierarchy may capture different facets. Additionally, most prior studies are cross-sectional and Western-based. The Saudi context with its young, predominantly female, nationally transitioning nursing workforce may produce distinct cultural dynamics that existing frameworks do not fully capture.

On Confirmatory Factor Analysis Results and Reliability

The results of the measurement model indicated satisfactory levels of measurement quality to allow the structural model to serve as a basis for inference. All of the standardized factor loadings for the standardized factors (.69−.79) are greater than.50 and therefore represent good measures of their respective latent constructs (1, 2, 3). Composites’ reliability coefficients (workplace culture = .92; burnout = .91; intent to stay = .85) are equal to or better than.70, which represents a reliable composite measure of its intended construct within the sample of Saudi nurses. In addition, the AVE values (range.51 −.54) meet the Fornell-Larcker criteria [27,28] indicating that each of the constructs explain at least half of the variance of the indicators. Furthermore, the discriminant validity indicates that the three constructs (workplace culture, burnout and intent to stay) are independent of one another because they meet the threshold for the Fornell-Larcker criteria [28], and HTMT ratios (<.85) [41,42]. As all three constructs are measured via self-report methods at one time, it is particularly important. Discriminant validity indicates that workplace culture, burnout, and intent to stay are distinct constructs in contrast to similar constructs in this context [43], which increases the likelihood that observed structural relationships reflect valid association as opposed to artifact caused by measurement error. Therefore, overall these measurement property results confirm that the measurement tools maintain their measurement properties when applied to a population of Saudi nurses. It is noted that the populations included in the original development of the measurement tools differ demographically and organizationally from the current population of interest. Validating contextual validity assures that mediation results are based upon measurements that are appropriate for this study’s context.

On Model fit Indices

The findings from this study emphasize the necessity of assessing both the structural and measurement validity of models examining nursing workforce trends. The structural equation model demonstrated mixed global fit: SRMR and RMSEA met criteria for good fit, whereas CFI and TLI were marginal (below the.95 threshold). This pattern partially aligns with simulation-based recommendations [30] that suggest a combination of SRMR < .08, CFI and TLI values near.95, and RMSEA near.06 indicates a well-defined relationship among variables. While RMSEA and SRMR met stringent criteria, CFI and TLI fell short of the.95 threshold, suggesting room for model improvement. For this Saudi nursing sample, the convergence of these multiple indicators confirms that the theoretical pathways between workplace culture, burnout, and intent to stay accurately reflect the data’s intercorrelations, providing a robust basis for validating the identified psychological mechanisms [44,45].

A primary strength of this analytical framework is the proactive examination of common method variance (CMV), a pervasive concern in cross-sectional, self-reported research [46]. While the Harman single-factor test indicated that CMV was not the predominant issue—as less than 50% of the total variance was attributed to a single factor—the test’s limited power necessitated a more rigorous unmeasured latent method factor (ULMF) analysis [46,47]. The ULMF results showed only a marginal increase in fit (ΔCFI = .01), which is considered insufficient to reach the criteria for substantial improvement.

The convergence of these diagnostic results establishes a credible empirical foundation for healthcare leadership in Saudi Arabia to pursue workplace culture as a high-impact retention strategy. Because the model fit is adequate—with good fit on absolute indices (RMSEA, SRMR) and marginal fit on incremental indices (CFI, TLI)—and both CMV diagnostics failed to find evidence of substantial bias. There is greater confidence that the structural path estimates reflect real-world organizational realities rather than statistical artifacts [48]. This confirms that the observed indirect effect is a genuine process—whereby poor work environments contribute to burnout and subsequent turnover—rather than an artificial construct created by methodological biases. Consequently, these results provide a scientifically supported rationale for prioritizing systemic cultural improvements to enhance nurse retention.

The Effects of Workplace Culture on Intention to Stay

The Structural Equation Modeling (SEM) examination in this study clearly shows workplace culture has a strong relationship with employee’s desire to remain in their position of employment. The finding that workplace culture relates to intent to stay through both direct and indirect pathways raises questions about the relative contribution of each mechanism. The direct pathway may reflect the influence of culture on job satisfaction, perceived organizational support, or relational identification factors not measured here but identified in prior retention research [49]. The indirect pathway through burnout suggests that culture also operates by modulating emotional exhaustion, which in turn colors nurses’ assessments of their future with the organization. As a result, creating a positive culture in the workplace is associated with lower levels of job-related stress.

In the structural model, the lowest ‘Intent to Stay’ outcome accounted for 38.9% of the variance. The modest variance explained in burnout (11.6%) suggests that other important antecedents not measured in this study—such as workload intensity, leadership quality, pay satisfaction, shift work patterns, and patient acuity—likely account for the majority of burnout variance. Future research should incorporate these job demands to more fully explain burnout etiology in this context. The small R² for burnout (11.6%) corresponds to the modest indirect effect (β = 0.09), indicating that the mediating pathway—while statistically significant—is practically narrow. Workplace culture’s influence on intent to stay operates primarily through the direct pathway (β = 0.48) rather than via burnout reduction. This suggests that cultural interventions may influence retention through multiple mechanisms beyond burnout alleviation alone, and that burnout-specific interventions are needed to capture the remaining 88.4% of burnout variance. Overall, these results are consistent with previous literature concerning the relationships between organizational climate, burnout, job satisfaction, and turnover intentions [50].

The present study provides contextual support for the Job Demands-Resources (JD-R) model within the Saudi nursing workforce. Rather than extending the theoretical framework, this study demonstrates that the JD-R model’s propositions hold in a specific cultural and occupational context characterized by rapid healthcare transformation, a young and predominantly female nursing workforce, and unique organizational structures. The findings are consistent with the JD-R model’s proposition that workplace culture functions as a contextual job resource that may buffer against job demands in this setting; however, given the cross-sectional design, the directional pathways implied by the model should be treated as hypothetical rather than confirmed consistent with the JD-R model’s original formulation [19]. This adds additional empirical support for the JD-R Model indicating that positive work conditions act as resources to lower the demands that lead to burnout and promote employee retention.

In addition, the SEM examination demonstrated that burnout acts as a partial mediator between workplace culture and the employee’s intent to stay. Thus, these cross-sectional findings suggest that more positive perceptions of the work environment co-occur with lower reported burnout, which in turn co-occurs with greater reported intent to stay. Burnout directly affects an individual’s decision making in terms of employment decisions [51]. The mediation role of burnout further emphasizes the need to prioritize employee mental health as a strategic way to affect organizational behaviors. While the traditional JD-R Model separates job demands that lead to burnout and organizational resources that stimulate engagement [52], the present study illustrates how workplace culture serves as an important resource that helps to alleviate work pressures and enhance employee resilience, thus affecting the intent to remain employed.

Nevertheless, while the SEM results provide convincing evidence, it is also reflective of the complexities that exist within previous literature. For example, relational organizational cultures actively buffer against a nurse’s intent to leave, whereas hierarchical workplace cultures, individual burnout, and workplace bullying act as severe, direct drivers of nurse turnover intentions [53]. This supports the idea that although a positive culture typically fosters employee retention, specific negative elements (i.e., incivility) can counteract those efforts. In addition, there have been other research that identified conflicting associations based on organizational structure. For example, a culture perceived as being strictly hierarchical may inadvertently increase employee turnover intentions in comparison to more innovative, less structured organizational environments [10]. Such contradictions illustrate that a “one-size-fits-all” approach to culture is likely to be inadequate. Therefore, additional research is needed to explore the complex nature of workplace culture and its differing implications across various sectors of industry.

The Demographic Covariates and the Primacy of Workplace Factors

The lack of significance of demographic variables implies that the mediated path from workplace culture to burnout to intent to remain does not represent a statistical artifact based on sample composition. If age had continued to be statistically significant, or if experience had been statistically significant, it would have suggested that the relationship between culture and burnout was merely a proxy for career stage effects; i.e., younger nurses were leaving because of initial career difficulties. However, the results show that the psychosocial work environment mediates between the two dimensions of demographics. Therefore, the JD-R model’s [19] idea that the resources provided by an employee’s work situation provide a stronger influence over withdrawal cognitions than do relatively stable personal characteristics has empirical validity. Furthermore, the results also suggest that burnout provides a universal explanation for turnover. The stability of the culture-burnout-retention pathway after accounting for demographic characteristics suggests that this association is not merely a proxy for career stage or nationality effects. This pattern supports the JD-R model’s emphasis on situational factors over individual differences in determining strain and withdrawal, though it does not rule out the possibility that demographic subgroups experience these processes with differing intensity. Because the work environment is a critical determinant of these individual behaviors and performance outcomes, organizations must prioritize creating supportive, well-resourced workplaces [54].

Although individual differences can exist among individuals based on their unique demographic backgrounds, they tend to account for a small amount of variance relative to the influence of the work environment. For example, although the Saudi population surveyed in this study represented a diverse group of nurses, the culture-burnout-retention pathway did not differ. As a result, the findings from previous studies support the conclusion that supportive cultures and burnout prevention are common approaches that may be applied across all types of nursing professionals [52] within the heterogenous Saudi nursing workforce. Practically speaking, hospital managers may wish to focus on making structural changes related to workplace culture, including providing employees with adequate supervision, effective communication systems and sufficient numbers of staff members working together. In doing so, hospital managers can avoid creating narrowly focused interventions designed for different demographic profiles and instead concentrate on addressing those factors common to virtually all employees (psychosocial aspects) in order to create maximum value out of limited retention resources available in a rapidly growing health care system.

While the JD-R model provides a useful organizing framework, it is not without limitations. The model assumes a unidirectional flow from resources and demands to strain to outcomes, yet longitudinal research suggests reciprocal effects: burnout may erode perceptions of workplace culture over time, and intent to stay may influence how employees appraise their work environment [17,18]. The cross-sectional design of this study cannot distinguish these directional alternatives. Additionally, the JD-R model treats burnout as the primary mediator, but other mechanisms such as organizational commitment, job satisfaction, or work engagement may operate in parallel or in sequence with burnout [55]. Conservation of Resources theory offers a complementary perspective: nurses who perceive their workplace as culturally supportive may experience less resource depletion and therefore be less likely to withdraw, regardless of whether they meet clinical criteria for burnout [6]. Future research could test multiple mediation models to determine whether burnout is the dominant pathway or one of several.

Study implications

These findings suggest several directions for future research and potential organizational consideration, pending confirmatory evidence from longitudinal or intervention studies. First, because workplace culture was associated with both lower burnout and greater intent to stay, hospitals might prioritize assessing cultural perceptions as part of retention monitoring. Second, the prominence of client-related burnout in this sample suggests that interventions addressing emotional demands of patient care such as debriefing protocols or staffing adjustments—could be evaluated for their impact on burnout and subsequent retention attitudes. Third, the 38.9% of variance in intent to stay explained by the model indicates that other unmeasured factors also warrant attention. Causal efficacy of any intervention cannot be established from cross-sectional self-report data; randomized controlled trials or quasi-experimental designs would be needed to determine whether culture-focused changes actually reduce turnover.

Study Limitations

While the results of this research are significant for understanding the relationship between workplace culture and burnout, several limitations must be considered. First, the cross-sectional design precludes causal inference, as data collection at a single point in time limits the ability to make definitive statements about cause-and-effect; future research should employ longitudinal or experimental designs to establish temporal precedence and examine how these variables evolve throughout a nurse’s career. Second, the reliance on self-reported data from an online survey may be susceptible to social desirability bias or common method variance, as respondents might report perceived expectations rather than their actual experiences. Third, although a proportionate stratified random sample was used across five hospitals in the Hail and Qassim regions, the findings may not generalize to the private healthcare sector or other geographic locations within Saudi Arabia. Furthermore, by excluding administrative staff and those on extended leave, the study results are specific to full-time clinical nurses rather than the entire nursing workforce. Fourth, the aggregation of nurses across clinical units which vary substantially in stressors, patient acuity, and resource availability leaves unit-level differences in the culture–burnout–retention pathway unexplored. Finally, the sample’s predominantly young, single, and relatively inexperienced demographic profile may shape retention attitudes and burnout vulnerability differently than in older or more senior cohorts, necessitating caution when generalizing these findings to more experienced nursing populations. The 41.1% male representation is higher than typical Saudi nursing workforce distributions, which likely reflects the inclusion of emergency, ICU, and operating room units settings that tend to have higher male nurse representation in Saudi Arabia. This enhances generalizability to these high-acuity settings but may limit generalizability to predominantly female units such as maternity or pediatric wards. Differential non-response by gender cannot be ruled out, though the stratified sampling approach should have mitigated systematic bias.

AI Use Declaration

We hereby declare that an AI-assisted editing tool, Paperpal, was utilized during the final preparation of this manuscript. Its use was strictly limited to English language editing, grammar correction, and improving text flow. The core scientific content, data interpretation, and intellectual substance of the manuscript were conceived and written entirely by the human authors. We maintain full responsibility for the originality and validity of the published work.